An open API service indexing awesome lists of open source software.

https://github.com/mariusei/schrodingers-claude

An operating manual for working with Claude — three principles (raskt, enkelt, pålitelig), a two-pass header workflow, and explicit discipline against LLM failure modes.
https://github.com/mariusei/schrodingers-claude

ai-coding-assistant claude claude-code llm manifesto

Last synced: about 1 month ago
JSON representation

An operating manual for working with Claude — three principles (raskt, enkelt, pålitelig), a two-pass header workflow, and explicit discipline against LLM failure modes.

Awesome Lists containing this project

README

          

# schrodingers-claude

An operating manual for working with Claude. The substance lives in
[CLAUDE.md](./CLAUDE.md); this README is a pointer.

Three principles, in Norwegian for precision:

- **Raskt** — fast to read, pivot, understand, and execute.
- **Enkelt** — each line earns its place; no bloat and no false simplification.
- **Pålitelig** — no assumptions, no TODOs, no hardcoded values, no untested code.

§5 of the manual is written for LLM agents specifically and names the failure
modes they are biased toward — premature collapse onto the obvious answer,
narrating expected outcomes, taking the shortest path, inventing certainty —
with explicit countermeasures. The repo name comes from this: hold the
solution space in superposition; collapse only on observation.

## Status

Distilled from working practice — the author runs a sibling of this manual on
macOS to good effect. Published as the first variant in what may become a
small comparative collection of CLAUDE.md manuals anchored on different
thinkers.

## Use

Drop `CLAUDE.md` into the root of your project; Claude Code loads it
automatically. Edit to taste — the principles are written specifically so
that disagreement is informative.